blob loss: instance imbalance aware loss functions for semantic segmentation
This addresses the challenge of detecting small instances in medical imaging, such as lesions for disease monitoring, though it is an incremental improvement over existing loss functions.
The paper tackled the problem of instance imbalance in semantic segmentation, where large instances dominate metrics like Dice coefficient, by proposing blob loss to improve detection of small instances, achieving a 5% F1 score improvement for MS lesions, 3% for liver tumors, and an average 2% for microscopy tasks.
Deep convolutional neural networks (CNN) have proven to be remarkably effective in semantic segmentation tasks. Most popular loss functions were introduced targeting improved volumetric scores, such as the Dice coefficient (DSC). By design, DSC can tackle class imbalance, however, it does not recognize instance imbalance within a class. As a result, a large foreground instance can dominate minor instances and still produce a satisfactory DSC. Nevertheless, detecting tiny instances is crucial for many applications, such as disease monitoring. For example, it is imperative to locate and surveil small-scale lesions in the follow-up of multiple sclerosis patients. We propose a novel family of loss functions, \emph{blob loss}, primarily aimed at maximizing instance-level detection metrics, such as F1 score and sensitivity. \emph{Blob loss} is designed for semantic segmentation problems where detecting multiple instances matters. We extensively evaluate a DSC-based \emph{blob loss} in five complex 3D semantic segmentation tasks featuring pronounced instance heterogeneity in terms of texture and morphology. Compared to soft Dice loss, we achieve 5% improvement for MS lesions, 3% improvement for liver tumor, and an average 2% improvement for microscopy segmentation tasks considering F1 score.